Nonlocal energies, such as fractional Sobolev seminorms, arise naturally in mathematical models involving long-range interactions. In this talk, we consider minimizers of such energies vanishing on a set of obstacles with random positions and sizes. This leads to what is commonly called the (bilateral) fractional obstacle problem. Our goal is to obtain a simplified description of the problem in the limit of many obstacles of small average size. Homogenization results of this type are known in both deterministic and random settings, but existing results in the random case typically rely on strong structural assumptions. I will present a new homogenization result that works under substantially weaker hypotheses. The obstacles may be generated by a broad class of spatial random models, allowing both their locations and their sizes to vary in a quite general way. In particular, obstacles are allowed to overlap, giving rise to a complex microstructure. The analysis identifies the limit of the energies and highlights how this reflects the probability distribution of the obstacles. The talk will be kept at an elementary level, starting with the classical local setting based on Poisson’s equation and then transitioning to the nonlocal framework, where the Laplace operator is replaced by its fractional counterpart.